Roles played by industries in promoting economic growth makes them the highest consumer of energy. The energy used by industries has proved difficult to forecast due to the nature of the independent variables. In recent years, artificial neural network (ANN) with multilayer perceptron (MLP) backpropagation function has been broadly utilized in the machine learning area. Most noticeable is in the field of forecasting. The MLP-ANN is applied to forecast South Africa’s energy consumption of five subsectors (basic chemicals, non-metallic minerals, basic iron and steel, basic non-ferrous metals and other manufacturing), with input variables in the form of activity, structure and intensity whereas energy consumed is the output from 1970 to 2016. In contrast with regression model, the results demonstrated a higher accuracy of industrial energy consumption in terms of statistical measures of performance.